Layer 21 · Neuron 2101 · suppressed

L21N2101-1

Word quirks:

“ dead”“保護”“ protected”“ protection”“ geschützt”“退”“的保护”“ Protected”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am Qwen, a language model developed by Anteriority (part of Alibaba's AI research arm). My mission is to be your unbiased assistant with a focus on safety, accuracy, and efficiency. I strive to provide clear, helpful responses while adhering to ethical guidelines and ensuring your privacy is protected. Whether you need help with complex reasoning, creative writing, or just a friendly chat, I'm here to support you responsibly! 😊

Nearest neighbors

Specimens with similar replies or word quirks. Similarity runs from −1 to 1; 1 is identical. These are leads, not shared mechanisms. How profiles are built.

By word quirks Profile recipe

Specimen	Similarity	Word quirks
1. L19N7566+1💤	0.91	“保护”“ protection”“ protect”
2. L25N6586+1💤	0.91	“保护”“ protection”“ Protection”
3. L26N995+1🕒	0.90	“ protected”“ protect”“保护”
4. L10N5998-1🕒	0.90	“ protected”“ protection”“ protective”
5. L24N4584+1💤	0.89	“保护”“ protection”“的保护”
6. L31N8291-1💥	0.89	“保护”“保護”“的保护”
7. L24N8540+1💤	0.89	“Protect”“保护”“ protecting”
8. L23N5608+1💤	0.89	“protect”“ protection”“ protective”
9. L24N8171-1💤	0.89	“保护”“ protection”“ Protection”
10. L28N8678-1💤	0.89	“保护”“protect”“保護”

Shadow specimen, apart from the ten: L21N2101+1💥, 0.04.

Cousins · specimens sharing a neuron

No specimen with replies shares this neuron yet.

Technical details

Quick scan: measure prediction shifts on fixed reference text. Short test: generate replies to a small set of prompts. Longer test: extend to a larger set. Full test: run the full prompt suite. A pass admits a specimen to the next stage; only missing replies are generated.

Layer 21, neuron 2101, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.59, top-1 agreement 0.71) is outside the band that feeds the short test (KL 1 to 2.5, agreement 0.5 to 0.65), so it stays at the quick scan unless someone queues it by hand. A hand pick goes straight to the full test. See Method for the tiers.

Quick scan · prediction measurements

Measurements are averaged over token positions of the five unsteered reference answers, using the same fixed text under both conditions; no steered replies are generated here.

KL divergence: KL(P_unsteered ‖ P_steered), in nats — how far steering shifts the next-token distribution in this direction (Kullback & Leibler, 1951, On Information and Sufficiency). Unsteered ≈ 0.001.
Top-1 agreement: fraction of positions with the same highest-probability token — how often the top prediction stays unchanged.
ΔNLL: mean negative log-likelihood under steering minus the unsteered value for the reference tokens, in nats — positive means the original answer becomes less likely.
Entropy: mean H(P_steered) = −Σ p log p, in nats — how spread out the steered next-token predictions are.

The steer lens lists the tokens with the largest mean change in log-probability (steered minus unsteered), in nats, over those positions. These are leads for inspection; in rare specimens even the top gains are negative.

Setting	KL divergence	Top-1 agreement	ΔNLL	Entropy	Steer lens
L21N2101+1💥 shadow specimen t0-v0.1	3.072	0.33	3.353	2.69	“uskan”“扔掉”“ dispens”“ disc”“ disp”“discard”“ discard”“deen”
L21N2101-1 t0-v0.1	0.585	0.71	0.747	1.27	Top boosted tokens ↑

Method

A specimen is a fixed steering configuration of Qwen3.5-4B: one MLP neuron or a pair, each with a signed dose. Unsteered is the control. L identifies the layer; N, the neuron.

Additive activation steering. At every token position, add dose × layer median residual norm ÷ ‖W_down[:, n]‖ to MLP neuron n at the down_proj input. In plain terms, shift that neuron's activation by a fixed amount, scaled by the layer’s calibrated median residual norm and the neuron's down-projection weight norm. Positive doses push up; negative doses push down. Dose is normalized intervention size, not a percentage or probability.